置件機優化問題討論已久，但大多數的研究只能適用於特定置件機型號或簡化問 題。而且在目前工業製造領域相對封閉的狀況下，優化所需的重要參數如儲料槽位置、 儲料槽間隔大小、材料移動速度限制或相機位置等只有置件機製造商能夠取得，因此 許多置件機通常會搭配專用程式，提供操作與優化功能。然而專用程式通常會有非常 多的限制，難以對優化的目標進行調整，也無法細部調整置件動作。因此本論文提出 一元優化架構，針對置件機優化中的儲料槽分配問題，使用基因演算法將專用程式的 優化成果再優化，並以 Panasonic NPM-W2 與專用程式 NPM-DGS 為例，設計客製化適 應函數與主觀函數以達到不同的優化目標。實驗成果顯示我們的架構除了達到額外優 化目標外，還能再改善 DGS 的優化結果 2.2% 至 38% 不等。 ;Mounter optimization has been discussed for a long time, while most of these works are aimed at certain mounter type or simplify the problem. Besides, in situation of that industrial manufacturing is relatively self-enclosed, most of these critical arguments required in optimization such as the position of feeder slots, the interval of nozzles in beams, the limited moving speed of each component and the position of the camera are held in the mounter manufacturer. Therefore, many mounters come with a dedicated program that provides operation and optimization function. However, there are many limitations to the dedicated program, which makes it hard to adjust the optimization target or modified mounting motion in detail. As a result, we propose a meta-optimization framework that focuses on feeder assignment problem which is a subproblem of mounter optimization. Also, we take Panasonic NPM-W2 along with dedicated NPM-DGS software as an example to demonstrate how to design a custom fitness/subjective function in order to fulfill different optimization targets. The experiment result shows that not only addition optimization targets are achieved, but also improved the result of DGS by decreasing total cycle time vary from 2.2% to 38%.
